A recent headline in Forbes declared, “AI won’t replace you, a human using AI will,” highlighting the transformative potential of artificial intelligence (AI) assistance in practice. In hospital medicine, one prominent implementation is the AI scribe, a tool designed to transcribe clinician-patient interactions in real time, populating the electronic health record with structured documentation.
Imagine walking into a patient room, microphone activated, speaking freely, with no notetaking required as the AI scribe listens in the background. Then, upon returning to the workstation, the encounter is already documented. Thus begins the era of the AI scribe, and its seductive promise: less typing, more talking.
In hospital medicine, the use of AI as a scribe is enticing, supported by data indicating that hospitalists spend approximately 17% of their time on direct patient care and 64% on indirect patient care, with 26% of the latter devoted to documentation.1 A similar distribution of time is seen among emergency department physicians, who spend 31% of their time on documentation.2 With hospitalists often allocating more time to reviewing electronic health records and documentation than to direct patient interaction, can an AI scribe help clinicians spend less time typing and more time gathering information from patients? In addition, according to the study “Where Did the Day Go?”, it was also observed that many hospitalists postpone documentation until later.3
AI-scribes offer a solution: real-time transcription of histories and physical exams, consultation notes, and progress notes. They promise to eliminate deferred documentation and enhance the time required for clinical documentation.
However, they may also introduce challenges in obtaining pertinent clinical information. If the AI is listening and recording, are we really listening? These apps promise time saved, but that time has to come from somewhere. And it may be coming from our cognitive presence at the bedside, listening, interpreting, and making meaning.
In recent discussions with colleagues regarding the use of AI scribes, I have noticed that users of AI transcription tools are often identifiable by the increased length of their histories of present illness. A pattern also extends to Epic Chat, where more extensive messages may suggest the involvement of an AI scribe. As a result, lengthy messages often require multiple readings to fully grasp the consultation or concern. Then I nostalgically remember being taught to present patients with one-liners. It forced clarity. Now I read notes and Epic Chats that span paragraphs.
While AI scribes may reduce time typing, I worry they might create barriers, not just between physicians and patients, but among clinicians ourselves. As technology integrates into practice, we must remember that faster is not always better and longer does not always mean more meaningful. Documentation should serve understanding, not obscure it.
Dr. Ortega-Sandoya
Dr. Ortega-Sandoya is a hospitalist at Memorial Hermann Sugar Land Hospital in Sugar Land, Texas.
References
1. Kim CS, et al. Hospitalist time usage and cyclicality: opportunities to improve efficiency. J Hosp Med. 2010;5(6):329-34. doi: 10.1002/jhm.613.
2. Füchtbauer LM, et al. Emergency department physicians spend only 25% of their working time on direct patient care. Dan Med J. 2013;60(1):A4558.
3. Tipping MD, et al. Where did the day go?—a time-motion study of hospitalists. J Hosp Med. 2010;5(6):323-8. doi: 10.1002/ jhm.790.